RIF1 promotes replication fork protection and efficient restart to maintain genome stability

Abstract

Homologous recombination (HR) and Fanconi Anemia (FA) pathway proteins in addition to their DNA repair functions, limit nuclease-mediated processing of stalled replication forks. However, the mechanism by which replication fork degradation results in genome instability is poorly understood. Here, we identify RIF1, a non-homologous end joining (NHEJ) factor, to be enriched at stalled replication forks. Rif1 knockout cells are proficient for recombination, but displayed degradation of reversed forks, which depends on DNA2 nuclease activity. Notably, RIF1-mediated protection of replication forks is independent of its function in NHEJ, but depends on its interaction with Protein Phosphatase 1. RIF1 deficiency delays fork restart and results in exposure of under-replicated DNA, which is the precursor of subsequent genomic instability. Our data implicate RIF1 to be an essential factor for replication fork protection, and uncover the mechanisms by which unprotected DNA replication forks can lead to genome instability in recombination-proficient conditions.

Fig. 1

RIF1 is recruited to the stalled replication forks. a Schematic representation of iPOND experiment. b Volcano plot showing the results for average fold-change to identify significantly upregulated proteins upon HU treatment based on H:L ratio in the SILAC experiment. The x-axis ('2Log Difference HU/NT) represents the fold upregulation. Data points in blue represent proteins that are upregulated >2-fold; RIF1 is indicated in red. c Bar graph showing fold upregulation of a selection of proteins upon HU treatment based on their SILAC H:L ratios (error bars represent standard deviation). d Representative micrographs showing co-localization of RIF1 (green) to sites of DNA replication as marked by EdU (red) in the presence or absence of HU in WT and 53bp1^−/− cells. Nucleus was stained with DAPI (blue). e Quantitation of d showing the percentage of cells, which show co-localization between EdU and RIF1(error bars represent standard deviation)

Fig. 2

Protection of reversed forks from degradation by RIF1. a Top panel: schematics of experimental conditions for fork progression in WT and Rif1^−/− MEFs. Cells were labeled with CldU (red) followed by IdU (green) as indicated. Representative DNA fibers for progression in WT and Rif1^−/− MEFs are shown below the schematic. Progression was measured by tract lengths of CldU (red) and IdU (green) in micrometers (μM). b Top panel: schematic for labeling cells in fork degradation assay. Representative pictures of normal and degraded fork are shown below the schematic. Cells were labeled with CldU followed by IdU and then subjected to replication stress with 4 mM HU for 3 h. Ratio of IdU to CldU tract length was plotted as readout for fork degradation. c, d Fork degradation assay in WT and RIF1-KO HAP1 cells (c) and between two different clones of WT, Rif1^−/−, and 53bp1^−/− MEF cell line (d). Experimental conditions were similar as in b. e Representative electron micrographs of normal fork (left) and reversed replication fork (right) observed on treatment with HU. The black arrow pointing to four-way junction at the replication fork indicates fork reversal (P, Parental, D, Daughter strand, R, Reversed arm). f Percentage of fork reversal in WT and Rif1^−/− MEFs treated with or without HU (4 mM) for 3 h. Numbers of analyzed molecules are indicated in parentheses. g WT and Rif1^−/− MEFs were transfected with siRad51 (100 nmols, 48 h) followed by labeling and treatment with 4 mM HU for 3 h. Fork degradation was determined in the presence and absence of RAD51. h Fork reversal frequencies observed with and without depletion of RAD51 in WT and Rif1^−/− MEFs under HU treatment. Numbers of analyzed molecules are indicated within parenthesis. Red bars in a, b, c, d, and g represent mean values from 125 fibers from each genotype under each condition. P-values were derived from Kruskal–Wallis ANOVA with Benjamini Hochberg (BH) post test except in c, where Mann–Whitney was used and in f and h, where unpaired t-test was done (ns, non-significant, ****P < 0.0001). All experiments were repeated three times with similar outcomes (Supplementary Data 2 and Supplementary Fig. 7a–e)

Fig. 3

DNA2 drives reversed fork degradation in RIF1-deficient cells. a Western blot analysis for the downregulation of MRE11 and DNA2 in WT and Rif1^−/− MEFs. WT and Rif1^−/− MEfs were transfected with either siControl or siRNAs smart pool against MRE11 and DNA2. Lysates made were probed with antibody against MRE11 and DNA2. Tubulin is used as loading control. b Ratio of IdU versus CldU in WT and Rif1^−/− MEFs upon HU treatment after downregulating Mre11 or DNA2 (a). c Ratio of IdU versus CldU in WT and RIF1-KO HAP1 cells upon HU treatment after inhibiting Mre11 and DNA2 using mirin and DNA2 inhibitor. d Electron microscopic analysis of percentage of reversed forks observed in WT and Rif1^−/− MEFs subjected to HU (4 mM) for 3 h, with or without DNA2 inhibitor. Numbers of analyzed molecules are indicated in parentheses. At least 125 readings were taken for b and c and the mean ratio is represented by red bar. P-values were derived from Kruskal–Wallis ANOVA with Benjamini Hochberg post test except in d, where unpaired t-test (ns, non-significant, ****P < 0.0001, **P = 0.0024) was carried out. Similar observation was made from three independent experiments (Supplementary Data 2 and Supplementary Fig. 7g, i)

Fig. 4

C-terminal region of RIF1 protects of reversed forks from degradation. a Schematic of full-length (FL) human RIF1 protein and deletion mutant constructs. Deleted region for each mutant is denoted by dotted line. b Western blot analysis for Rif1^−/− MEFs transfected with mutant construct of human RIF1. Lysates were probed with antibody against GFP. XPD was used as loading control. Expression of mutant protein is visualized as distinct bands in range of 198 kD to 310 kD, which is missing in mock-transfected samples. c DNA fiber assay to assess the rescue of Fork degradation in Rif1^−/− MEFs transfected with RIF1 mutant constructs (for 48 h) upon treatment with 4 mM HU for 3 h. d Percentage of fork reversal in Rif1^−/− MEFs transfected with different mutant constructs of human RIF1 and subsequent treatment with HU for 3 h (4 mM). Numbers of analyzed molecules are indicated in parentheses. e DNA fiber assay to determine the extent of fork degradation in WT and Rif1^−/− MEFs upon siRNA-mediated downregulation of PP1. f Percentage of reversed forks observed in WT and Rif1^−/− MEFs treated with 4 mM HU for 3 h with or without inhibiting PP1. Number of molecules analyzed are indicated within the parenthesis. At least 125 readings were taken for c and e and the mean values are represented by red bar. P-values were derived from Kruskal–Wallis ANOVA with Benjamini Hochberg post test for c and e and from unpaired t-test for d (ns, non-significant, ***P = 0.0009, **P = 0.0025) and f (ns, non-significant, ***P = 0.0003, **P = 0.0026). All the experiments were repeated for three times with similar outcomes (Supplementary Data 2 and Supplementary Fig. 8a, b). g DNA2 is hyper phosphorylated in Rif1^−/− MEFs during replication stress. Top panel: level of DNA2 in nuclear extracts from WT and Rif1^−/− MEFs before and after treatment with HU alone or in combination with PP1 inhibitor treatment (tautomycetin 225 nM for 2 h). Western blots were performed with antibody against DNA2 antibody. Histone H3 was used as loading control. Bottom panel: Immunoprecipitations were carried out with anti-DNA2 antibody or the corresponding IgG and were probed with p-(S/T) antibody

Fig. 5

Delayed fork restart and genomic instability observed upon RIF1 deficiency. a PFGE analysis for DSBs in WT and Rif1^−/− MEFs with and without treatment with HU for 3 h. WT MEFs treated with IR (15 Gy) was taken as positive control. b Quantification of experiment (a), an integration of three independent experiments showing DSB levels relative to WT untreated (NT), (ns, not-significant, from unpaired t-test). c Representative images for analysis of genomic instability analysis by metaphase spread in WT and Rif1^−/− MEFs upon HU and Cisplatin treatment. d Quantitation of chromosomal aberrations in c. Sixty metaphase fields per conditions were analyzed and three independent experiments were carried out. P-value was calculated by unpaired t-test (***P ≤ 0.0001). e–f Images for clonogenic survival assay in WT and Rif1^−/− MEFs treated with different concentrations of HU (e) and Cisplatin (f) after which the drugs were washed off and the cells were allowed to grow for 8 days. Adjoining graphs show the data from three independent experiments. Error bars represent s.e.m. g Schematics of fork restart assay by DNA fibers and representative images for normal restart, delayed restart and stalled fork upon release from HU treatment. h Quantitation for restart assay in g. Tract lengths of IdU and CldU were quantified in WT and Rif1^−/− MEFs upon restart after treatment with 1 mM HU for 1 h from 125 fibers per sample. Red and green bars indicate mean CldU and IdU tract length. P-values were derived from Kruskal–Wallis ANOVA with Benjamini Hochberg post test. All experiments were repeated three times (Supplementary Data 2 and Supplementary Fig. 8d)

Fig. 6

Restart defects are a consequence of fork degradation in RIF1-deficient cells. a PFGE in WT and Rif1^−/− MEFs with and without treatment with HU for 3 h and 15 h recovery after treatment. b Quantification of experiment (a), from three independent experiments showing DSB levels relative to WT untreated (NT), (ns, not-significant, **P = 0.0019, unpaired t-test). c, d Electron micrographs of ssDNA at the fork (c), and behind the fork (d), 30 min after release from HU treatment. White arrows represent ssDNA at the forks and black arrows in d, represent ssDNA gaps behind the forks e Analysis of ssDNA at forks upon restart in WT and Rif1^−/− MEFs in presence or absence of DNA2 inhibitor. Red bar represents mean, P-value was derived from Kruskal–Wallis ANOVA with Benjamini Hochberg post test. f Analysis of internal gaps behind forks upon restart in WT and Rif1^−/− MEFs in the presence or absence of DNA2 inhibitor and HU. Graph represents mean and SD from three independent experiments. Chi-square test of trends was done to assess significance of internal ssDNA gaps between WT and Rif1^−/− MEFs (ns, non-significant, ****P < 0.0001). Numbers of analyzed molecules are indicated within parenthesis for e, f. g Top: schematics for restart assay by fibers upon DNA2 inhibition. Bottom: Tract lengths of IdU and CldU were quantified in WT and Rif1^−/− MEFs upon restart after treatment with 1 mM HU for 1 h in the presence or absence of DNA2i. h Top: schematics for fiber restart assay upon transfection of hRIF1 deletion mutant constructs in Rif1^−/− MEFs. Bottom: Quantification of IdU tracts in Rif1^−/− MEFs upon restart after treatment with 1 mM HU for 1 h in the presence or absence of hRIF1 deletion constructs. Red and green bars in g and h represents mean CldU and IdU tract length, P-values were obtained from Kruskal–Wallis ANOVA with Benjamini Hochberg post test for FDR. All experiments were repeated thrice (Supplementary Data 2 and Supplementary Fig. 8h–i). i Survival assay in Rif1^−/− MEFs complemented with hRIF1-FL, Del-HEAT, Del-CI, Del-CII constructs of hRIF1 and treated with different concentrations of HU and Cisplatin. Data represented from three independent experiments, error bars represent s.e.m

Fig. 7

Model for role of RIF1 in fork protection and genome stability. a Replication stress in cells results in replication fork reversal to stabilize stalled replication forks. b Fork reversal results in the recruitment of RIF1 probably through its C-terminal domain, which has cruciform structure binding properties. Binding of RIF1 to reversed forks stabilizes them by recruitment of PP1, which brings about de-phosphorylation of DNA2 and thereby limits access of DNA2 nuclease to these forks and prevents fork degradation. This allows for normal restart of reversed forks probably through RECQ1-mediated branch migration of these reversed forks resulting in prevention of genome instability and cellular viability upon replication stress. c In contrast, absence of RIF1 results in DNA2-mediated degradation of reversed forks. In the absence of the preferred substrate (four-way junctions), RECQ1 is unable to bind. Forks are therefore aberrantly restarted which results in exposure of under-replicated DNA in the form of ssDNA gaps behind the forks. These ssDNA gaps become a source of genome instability and DSBs later during the cell cycle in G2/M phases resulting in sensitivity to replication stress-inducing agents